[molpro-user] MS-MR-CASPT2: gradient calculation changes the wavefunction - which is the "right" one?

[Sve N]

Dear molpro-users,
again I have a question concerning wavefunctions. I'd like to use a MS-MR-CASPT2 wavefunction in the DDR-Routine, but as soon as I perform a gradient calculation, the wavefunction, stored via "save" changes. I'm not sure if this is intended, since also all information in the result-block changes with it, and I don't know, which wavefunction is the correct one. I'll put an example calculation at the end of this mail, that shows the different results after the CP-CASPT2-calculation with the same MS-CASPT2 analysis block in front of it. An additional line "save,6100.2;" would cause different results in a subsequent DDR-calculation, which is my ultimate problem, but I think it is connected with this, shorter to demonstrate effect.
The second question, if someone knows, which wavefunction is the correct one, would be how to store it, without being forced to either do, or not to do a gradient calculation, which would result in a waste of calculation time in both cases.
Thanks in advance, Sven

P.S.: The example calculation (molpro 2006):
__________________________________________
***Ethen
memory,100,m

BASIS
DEFAULT=6-31G*
END;

geomtyp=xyz
geometry={
Angstrom,NOORIENT,nosym;
6

 C     0.822920    -0.033882     0.015023
C    -0.876435    -0.065389    -0.030613
H     1.739927    -0.708451     0.415439
H     1.468778     0.797877    -0.390777
H    -1.257397     1.044249     0.221520
H    -0.888192    -1.109452    -0.343283
}

hf;

{multi;
Maxit,30;
occ,9;
closed,7;
state,3;
canorb,2140.2;
}

{ci;
occ,9
closed,7;
noexc;
state,3;
orbital,2140.2
}

{rs2,shift=0.3,maxiti=250,MIX=3;
state,3;
occ,9;
closed,7;
orbital,2140.2;
}

{rs2,shift=0.3,maxiti=250,MIX=3,root=1;
state,3;
occ,9;
closed,7;
orbital,2140.2;
}

FORCE
---
________________________________________
Output:

[...]
1PROGRAM * RS2 (Multireference RS Perturbation Theory)     Authors: H.-J. Werner (1993), P. Celani (1998)
[...]
 Heff Half Sum symmetrized
                   1             2             3
         1   -78.1315905    -0.0099545    -0.0223860
         2    -0.0099545   -77.9659878     0.0027797
         3    -0.0223860     0.0027797   -77.9051637

 MS-CASPT2 energies

 STATE       ENERGY     MIXING COEFFICIENTS
    1    -78.1343342    0.9937    0.0572    0.0964
    2    -77.9656582   -0.0511    0.9966   -0.0647
    3    -77.9027495   -0.0997    0.0594    0.9932

 *******************************************
 * End of MS-CASPT2 analysis               *
 * Properties not yet implemented          *
 *******************************************
[...]
 !RSPT2 STATE 1.1 ENERGY              -78.131590484966
 !RSPT2 STATE 1.1 DIPOLE MOMENT        -0.00743787     0.06875973    -0.03829846
[...]
1PROGRAM * RS2 (Multireference RS Perturbation Theory)     Authors: H.-J. Werner (1993), P. Celani (1998)
[...]
 Heff Half Sum symmetrized
                   1             2             3
         1   -78.1315905    -0.0099545    -0.0223860
         2    -0.0099545   -77.9659878     0.0027797
         3    -0.0223860     0.0027797   -77.9051637

 MS-CASPT2 energies

 STATE       ENERGY     MIXING COEFFICIENTS
    1    -78.1343342    0.9937    0.0572    0.0964
    2    -77.9656582   -0.0511    0.9966   -0.0647
    3    -77.9027495   -0.0997    0.0594    0.9932

 Norm of Psi_mix recalculated     1.10323160

 *******************************************
 * End of MS-CASPT2 analysis               *
 * Properties not yet implemented          *
 *******************************************

 Now solve CP-CASPT2 equations
[...]
 !RSPT2 STATE 1.1 ENERGY              -78.134334262310
 !RSPT2 STATE 1.1 DIPOLE MOMENT        -0.01240846     0.06598818    -0.03582002
[...]

 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20110926/d59a1344/attachment.html>